Self-Organization in Autonomous Sensor/Actuator Networks [SelfOrg] Dr.-Ing. Falko Dressler Computer Networks and Communication Systems Department of Computer Sciences University of Erlangen-Nürnberg http://www7.informatik.uni-erlangen.de/~dressler/ dressler@informatik.uni-erlangen.de

Overview Self-Organization Introduction; system management and control; principles and characteristics; natural self-organization; methods and techniques Networking Aspects: Ad Hoc and Sensor Networks Ad hoc and sensor networks; self-organization in sensor networks; evaluation criteria; medium access control; ad hoc routing; data-centric networking; clustering Coordination and Control: Sensor and Actor Networks Sensor and actor networks; coordination and synchronization; in-network operation and control; task and resource allocation Bio-inspired Networking Swarm intelligence; artificial immune system; cellular signaling pathways

Mobile Ad Hoc and Sensor Networks Mobile Ad Hoc Networks (MANET) Wireless Sensor Networks (WSN)

Infrastructure-based Wireless Networks Typical wireless network are based on infrastructure E.g., GSM, UMTS, WLAN, … Base stations connected to a wired backbone network Mobile entities communicate wirelessly to these base stations Traffic between different mobile entities is relayed by base stations and wired backbone Mobility is supported by switching from one base station to another Backbone infrastructure required for administrative tasks IP backbone Further networks Gateways Server Router

Infrastructure-based Wireless Networks – Limitations? What if … No infrastructure is available? – E.g., in disaster areas It is too expensive/inconvenient to set up? – E.g., in remote, large construction sites There is no time to set it up? – E.g., in military operations

Possible Applications for Infrastructure-free Networks Factory floor automation Disaster recovery Car-to-car communication Finding out empty parking lots in a city, without asking a server Search-and-rescue in an avalanche Personal area networking (watch, glasses, PDA, medical appliance, …) Military networking: Tanks, soldiers, … …

Further Applications Collaborative and Distributed Computing Temporary communication infrastructure Quick communication with minimal configuration among a group of people Examples A group of researchers who want to share their research findings during a conference A lecturer distributing notes to a class on the fly Emergency operations Rescue, crowd control, and commando operations Constraints Self-configuration with minimal overhead Independency of fixed or central infrastructure Freedom and flexibility of mobility Unavailability of conventional communication infrastructure

Solution: (Wireless) Ad Hoc Networks Try to construct a network without infrastructure, using networking abilities of the participants This is an ad hoc network – a network constructed “on demand”, “for a special purpose” Simplest example: Laptops in a conference room – a single-hop ad hoc network

Limited range: Multi-hopping For many scenarios, communication with peers outside immediate communication range is required Direct communication limited because of distance, obstacles, … Solution: multi-hop network ?

Wireless Mesh Networks Alternate communication infrastructure for mobile or fixed nodes/users Independence of spectrum reuse constraints and the requirements of network planning of cellular networks Mesh topology provides many alternate data paths Quick reconfiguration when the existing path fails due to node failures Most economical data transfer capability coupled with the freedom of mobility

Ad Hoc Networks vs. Infrastructure-based Networks Prerequisites Pre-deployed infrastructure, e.g. routers, switches, base stations, servers None Node properties End system only Duality of end system and network functions Connections Wired or wireless Usually wireless Topology Outlined by the pre-deployed infrastructure Self-organized topology maintained by the nodes Network functions Provided by the infrastructure Distributed to all participating nodes

Mobility: Suitable, Adaptive Protocols In many (not all!) ad hoc network applications, participants move around In cellular network: simply hand over to another base station In mobile ad hoc networks (MANET): Mobility changes neighborhood relationship Must be compensated for E.g., routes in the network have to be changed Complicated by scale Large number of such nodes difficult to support

MANET (Mobile Ad Hoc Network) Active IETF working group Standardization of IP routing protocol functionality suitable for wireless routing application within both, static and dynamic topologies Approaches are intended to be relatively lightweight in nature, suitable for multiple hardware and wireless environments, where MANETs are deployed at the edges of an IP infrastructure Support for hybrid mesh infrastructures (e.g., a mixture of fixed and mobile routers)

Battery-operated Devices: Energy-efficient Operation Often (not always!), participants in an ad hoc network draw energy from batteries Desirable: long run time for Individual devices Network as a whole Energy-efficient networking protocols E.g., use multi-hop routes with low energy consumption (energy/bit) E.g., take available battery capacity of devices into account How to resolve conflicts between different optimizations?

Problems/Challenges for (Mobile) Ad Hoc Networks Without a central infrastructure, things become much more difficult Lack of central entity for organization available Limited range of wireless communication Mobility of participants Battery-operated entities Without a central entity (like a base station), participants must organize themselves into a network  Self-organization Pertains to (among others) Medium access control – no base station can assign transmission resources, must be decided in a distributed fashion Finding a route from one participant to another

Wireless Sensor Networks Participants in the previous examples were devices close to a human user, interacting with humans Alternative concept: Instead of focusing interaction on humans, focus on interacting with environment Network is embedded in environment Nodes in the network are equipped with sensing and actuation to measure/influence environment Nodes process information and communicate it wirelessly Wireless sensor networks (WSN)

Wireless Sensor Networks Multiple roles can be distinguished Sensors – measure physical phenomena, sources of measurement data Base stations – analyze and post-process data, sinks for measurement data Actuators – perform actuation in response to received data Processing elements – pre-processing of transmitted data base station sensor node

Composition of Sensor Nodes – Hardware Processor (and memory) E.g., Atmel ATmega128 microcontroller, 16 MHz, 128 kByte flash Radio transceiver E.g., Chipcon CC1000 (315/433/868/915 MHz), CC2400 (2.4 GHz) Battery Possibly in combination with energy harvesting Sensors Light, temperature, motion, … Sensor 1 Micro controller Radio transceiver … Sensor n Memory Storage Battery

Composition of Sensor Nodes – Software Event-driven operating principle E.g., TinyOS System component Event (Sensor) System function New event (Timer) … Event (Timer) … … … System function New event (Data packet) Event (Transceiver) Event handler

Communication in WSN MAC Network Transport Application Energy-efficiency Network Address-based routing Data-centric routing Transport Data aggregation Application Push vs. pull Application layer Task management plane Transport layer Mobility management plane Power management plane Network layer MAC layer Physical layer

Communication in WSN Push vs. pull base station source 2 source Request (“pull”) source 1 source 2 Periodic transmission (“push”) base station source source 2 Transmission (“pull”)

Deployment Options for WSN How are sensor nodes deployed in their environment? Dropped from aircraft  Random deployment Usually uniform random distribution for nodes over finite area is assumed Is that a likely proposition? Well planned, fixed  Regular deployment E.g., in preventive maintenance or similar Not necessarily geometric structure, but that is often a convenient assumption Mobile sensor nodes Can move to compensate for deployment shortcomings Can be passively moved around by some external force (wind, water) Can actively seek out “interesting” areas

Deployment Options for WSN Evaluation criteria? Coverage! Radio coverage, i.e. communication related Sensor coverage, i.e. application related

MANET vs. WSN Many commonalities Many differences Self-organization, energy efficiency, (often) wireless multi-hop Many differences Applications, equipment: MANETs more powerful (read: expensive) equipment assumed, often “human in the loop”-type applications, higher data rates, more resources Application-specific: WSNs depend much stronger on application specifics; MANETs comparably uniform Environment interaction: core of WSN, absent in MANET Scale: WSN might be much larger (although contestable) Energy: WSN tighter requirements, maintenance issues Dependability/QoS: in WSN, individual node may be dispensable (network matters), QoS different because of different applications Data centric vs. id-centric networking Mobility: different mobility patterns like (in WSN, sinks might be mobile while nodes are usually static)

WSN Application Examples Emergency operations Drop sensor nodes from an aircraft over a wildfire Each node measures temperature Derive a “temperature map” Habitat monitoring Use sensor nodes to observe wildlife E.g., Great Duck Island, ZebraNet Precision agriculture Bring out fertilizer/pesticides/irrigation only where needed Logistics Equip goods (parcels, containers) with a sensor node Track their whereabouts – total asset management Note: passive readout might suffice – compare RFIDs

WSN Application Scenarios Home automation and health care Smart environment (smart sensor nodes and actuators in appliances learn to provide needed service) Post-operative or intensive care (telemonitoring of physiologic data) Long-term surveillance of chronically ill patients or the elderly (tracking and monitoring)

Operation and Maintenance Type of service of WSN Not simply moving bits like another network Rather: provide answers (not just numbers) Issues like geographic scoping are natural requirements, absent from other networks Feasible and/or practical to maintain sensor nodes? E.g., to replace batteries? Or: unattended operation? Impossible but not relevant? Mission lifetime might be very small Energy supply? Limited from point of deployment? Some form of recharging, energy scavenging from environment? E.g., solar cells

Research Objectives Network lifetime The network should fulfill its task as long as possible – definition depends on application Lifetime of individual nodes relatively unimportant But often treated equivalently Maintainability and fault tolerance WSN has to adapt to changes, self-monitoring, adapt operation Incorporate possible additional resources, e.g., newly deployed nodes Be robust against node failures (running out of energy, physical destruction, …) In-network processing Again, the network should fulfill a given task on behalf of an external user Move necessary computations into the network  reduction of communication costs, speedup of operations Definition of network lifetime as a homework?

Research Objectives Quality of service Software management Traditional QoS metrics do not apply Still, service of WSN must be “good”: Right answers at the right time Software management Programming and re-programming of sensor nodes according to the current application demands Debugging of distributed heterogeneous sensor nodes? From ZebraNet: “how to reboot a zebra?” Definition of network lifetime as a homework?

Self-Organization in Sensor Networks

Principles and properties Self-organizing networks Conventional networks Local state Networking functions for global connectivity and efficient resource usage Global state (globally optimized system behavior) Neighbor information Probabilistic methods Implicit coordination Explicit coordination

Self-Organization in WSN Objectives Scalability – Management overhead for coordination, support for “unlimited?” number of nodes Lifetime – Application dependent description of the service quality including delays and availability Categorization in two dimensions Horizontal, i.e. according to the necessary state information Vertical , i.e. according to the network layer

Horizontal dimension Location information Neighborhood information Absolute or relative position, affiliation to a group of nodes Usually requires multi-hop communication Neighborhood information Direct neighborhood, based on local broadcasts Local state Local system state, environmental factors Probabilistic algorithms No state information required, stochastic processes - table-driven routing - medium access control - data centric routing - task allocation collision avoidance gossiping - topology control - clustering location information neighborhood information local state probabilistic algorithms

Vertical dimension MAC layer Network layer Application layer Medium access, local communication Network layer Topology control, routing tables, data-centric communication Application layer Coordination and control, application dependent requirements (coverage, lifetime) Control plane (e.g. mobility management) Application layer Transport layer Cross-layer optimization (e.g. energy control) Network layer MAC layer Physical layer

Mapping of Primary Self-Organization Techniques Location information Neighborhood information Local state Probabilistic methods Feedback loops Feedback is provided by observing and evaluating system parameters; this can be done either by local means (sensor readings) or with external help of neighboring systems Interactions Information exchange among remote nodes using routing techniques Local interaction among direct neighbors within their wireless communication range Interactions with the environment or indirect interactions with other nodes using environmental changes (stigmergy) Probabilistic techniques Randomness is often exploited to prevent unwanted synchronization effects, e.g. for retrial attempts Stochastic methods

Further Studies Medium access control (MAC) Ad hoc routing Problems and solutions Case studies (S-MAC, PCM) Ad hoc routing Classification Principles of routing protocols Optimized route stability Address allocation techniques Data-centric networking Flooding, gossiping, and optimizations Agent-based techniques Directed diffusion Clustering Principles and techniques Case studies (LEACH, HEED)

Summary (what do I need to know) Principles of ad hoc and sensor networks Commonalities Differences Capabilities and working behavior of WSN Node hardware and software Communication principles Self-organization in WSN Two-dimensions Mapping to “classical” self-organization techniques

References I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, "Wireless sensor networks: a survey," Computer Networks, vol. 38, pp. 393-422, 2002. D. Culler, D. Estrin, and M. B. Srivastava, "Overview of Sensor Networks," Computer, vol. 37 (8), pp. 41-49, August 2004. I. Dietrich and F. Dressler, "On the Lifetime of Wireless Sensor Networks," University of Erlangen, Dept. of Computer Science 7, Technical Report 04/06, December 2006. F. Dressler, "Self-Organization in Ad Hoc Networks: Overview and Classification," University of Erlangen, Dept. of Computer Science 7, Technical Report 02/06, March 2006. H. Karl and A. Willig, Protocols and Architectures for Wireless Sensor Networks, Wiley, 2005. C. Prehofer and C. Bettstetter, "Self-Organization in Communication Networks: Principles and Design Paradigms," IEEE Communications Magazine, vol. 43 (7), pp. 78-85, July 2005. C. S. Raghavendra, K. M. Sivalingam, and T. Znati, Wireless Sensor Networks. Boston, Kluwer Academic Publishers, 2004. H. Zhang and J. C. Hou, "Maintaining Sensing Coverage and Connectivity in Large Sensor Networks," Wireless Ad Hoc and Sensor Networks: An International Journal, vol. 1 (1-2), pp. 89-123, January 2005.